Author :
Alan Kan
Every few weeks, someone asks me whether they should still learn UX design.
The headlines out there are scary. Figma's First Draft turns a sentence into a working layout. Google's Stitch builds an interface, and the front-end code, from a prompt or a photo of a whiteboard. Describe a screen to Claude and it'll build one in front of you, in Artifacts, while you watch.
My honest answer comes in two halves.
The Hard Part
Getting in has become harder. The production tasks juniors used to learn on, such as the wireframes, the layouts, the "just build the thing" work, are now the things AI does in seconds, for free. That hits newcomers hardest, and a lot of people are reading it as a sign the field is closing. I don't agree with that. The field is changing, and that's a reason to bring newcomers in differently, not to walk away from it.
Now the Good News
The tools got very good at producing screens. They got nowhere near as good at knowing which screen is the right one, why it's right, or what to do when the thing falls over. That gap is human for a reason: AI works from patterns in its training data, not from sitting with a real user in a real situation and noticing what they actually do, struggle with, or quietly want. It doesn't know your customers, your context, or the messy reality the product has to survive in. Output is cheap now. Discernment isn't, and it's what the market has quietly started paying for. After years of being told UX was just making things pretty, that's a welcome shift.
The tools got very good at producing screens. They got nowhere near as good at knowing which screen is the right one.
There's a bigger reason to be optimistic, though. Once anyone can generate software from a few sentences, the world fills up with software (most of it mediocre). More products, shipped faster, by more people, and every one of them needs somebody who can make it usable by an actual human. All that cheap production creates the work; it doesn't remove it.
The work is also genuinely harder than it used to be. AI gets things wrong, often while sounding completely sure of itself, so designers now have to build for a system that might hand the user nonsense. When an agent starts doing things on someone's behalf, you have to design how they see what it did and how they undo it. Those are real problems, and the tools generating all the pretty screens don't touch them. That's the UX job now.
What the job asks for now
The work has moved on from producing deliverables, because that's the part AI took. What holds its value is judgment, research, and being able to frame a design decision in terms a business actually cares about. Fluency with the AI tools belongs in there too — as part of the craft rather than a bolt-on — knowing how to drive them, and knowing when to set them aside and bring a human back in. The people who do well from here are the ones who turn up understanding the problem, not just the software. That's the bit an employer can't swap out for a prompt.
So how do newcomers get trained and get jobs in UX design?
If the old entry-level work is gone, the old route in goes with it. You can't spend a few years doing production grunt work and pick up judgment along the way, because the grunt work is what got automated. So the judgment has to come first instead of last. And the catch with judgment is that you can't really get it from a classroom, or from another polished pet project. It comes from experience of real work: real clients, real constraints, real users, and the discomfort of making a call and living with how it plays out. A made-up brief lets you off the hook; a real one with a real customer doesn't.
A made-up brief lets you off the hook; a real one doesn't.
Newcomers need that exposure - sitting with actual users, shipping something that matters to someone, defending a decision to a person who'll push back - well before they'd have earned it under the old model. Fluency with the AI tools is part of it too, but it's the cheap part to pick up. Someone who can show how they think, and why they ruled the other options out, is far more employable now than someone with a folder of tidy mockups. The screens were never the hard part; the thinking behind them was, and you only really learn that by doing the work.
Where I land
The future of this field belongs to people who can take complex, fallible AI systems and make them genuinely usable (that goes way beyond just chat interfaces) and trustworthy for real human beings. That's harder than UX has ever been, and it matters more than it ever has, because there has never been this much software, built this fast, with this little thought for the person on the other end of it.
AI didn't replace designers, it raised the bar for getting in. That's good news for anyone who can show real experience, not just a tidy portfolio. Done properly, UX skills might matter more now than they ever have.
Written by me with assistance from AI